newclassd: the better class d amp?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I wonder how something with pre-filter feedback can be considered the best. Trebble output is uncontrolled, speaker dependent. Output impedance rises at 6dB/oct (thus damping falls at the same rate). Output filter non-linearity (and any cross-talk) is uncontrolled too.

On the other hand, when you feel good or excited about something new or just not tired, everything sounds better, more detailed. For example, I can get very good and very bad impressions about exactly the same system at different times of the day, that's why I have never trusted subjective reviews. You are evaluating many feelings not related to sound as much as the ones related to sound.

Sorry Eva but you have no clue what these Tripath chips can do.
Listing a bunch of so called problems will not change the fact that
my Accuphase E303, SKA100, AKSA 50N EL84 tube amp, all could not
even come close to my modded Sure 2X100 bords.
They are so natural, detailed, airy and throw a sound-field to die for.
I doubt that I will ever use any other amp than those. I might try
a truepath in the future, but that can take a while, because I am utterly
satisfied at the moment.

Greets,
Klaus
 
Darkfenriz:

I think almost every classd amplifier uses 2nd order filtering, this one is no different. However the filter function does not merely consist of an inductor and a capacitor.

The special design of the output inductor with a very low parallel capacitance, blocks high frequency components from the output stage from passing to the speaker wires. The residual carrier is approximately 4 Vpp at specified power.

AP2:

The drivers are built to run continously at 750 KHz, or even higher, but higher switching frequency does not improve the characteristics of the amplifier. Apart from the carrier residuals. Normally we get a good compromise at 500 kHz, which is chosen as the optimal switching frequency based on this.

Radian:

I would agree with you, at this point Class D technology has surpassed the performance of even very good "old school" linear technologies. I didn't hear the Sure boards yet, so i couldn't comment on those. The data don't look like high end, but that doesn't mean the amp can't sound good.
 
I have read the comparsion test of various ClassD modules about
http://audiodesignguide.com/ClassD/classD.html

Where are to find simplefield and detailled schematics of the newclassD modules and papers about the theory of operation and design advices?
About
http://www.diyaudio.com/forums/clas...ic-about-rev-engin-not-complete-finished.html
you will find a fragment of the schematic from the smallest ICEPOWER module. Very interesting is by the ICEPOWER module the avoiding of custom made IC's and therefore such topologies actually very appropriated for diy, pSpice simulation, evaluation and realizing own ideas.

This seems also to be by the newclassD modules (see attachement)

Theoretical background concerning the ICEPOWER topology there is to find a lot.

Are there between the newclassD modules and the ICEPOWER modules similarities in the circuit topology?

A second question: is it possible, to replace integrated amplifiers with an newclassD module against the ICEPOWER 200AC Rev.D (I have now a MONRIO integrated amp for repair - the used ICEPOWER modules complete damaged, so that I must replace most semiconductors and SMD inductors) ?
 

Attachments

  • newClassD soldersite.pdf
    96.4 KB · Views: 434
  • newclassD topview.pdf
    151 KB · Views: 370
Last edited:
Ok, 4Vpp@500kHz residual from pre-filter 120Vpp PWM square (~30dB attenuation), filter of second order (-12dB/oct) tuned to -3dB@200kHz, right?
Since this does not make sense I ask where's the trick,
BD modulation?
multiphase operation?
...?
none of the above?

Edit:
Keep in mind that basic sampling theory limits the bandwidth to half the switching frequency ;)
 
Last edited:
Ok, 4Vpp@500kHz residual from pre-filter 120Vpp PWM square (~30dB attenuation), filter of second order (-12dB/oct) tuned to -3dB@200kHz, right?
Since this does not make sense I ask where's the trick,
BD modulation?
multiphase operation?
...?
none of the above?

Edit:
Keep in mind that basic sampling theory limits the bandwidth to half the switching frequency ;)

Darkfenriz,
Have some mercy on Lars, just buy his module and measure its specs as per your liking............Lars shouldn't be subjected to this much mental torture at least.;)

BTW:I still miss your old cat avtar :)
 
Last edited:
Ok, 4Vpp@500kHz residual from pre-filter 120Vpp PWM square (~30dB attenuation), filter of second order (-12dB/oct) tuned to -3dB@200kHz, right?
Since this does not make sense I ask where's the trick,
BD modulation?
multiphase operation?
...?
none of the above?

Edit:
Keep in mind that basic sampling theory limits the bandwidth to half the switching frequency ;)

Dark Fenriz: I don't wish to try to teach you about basic LC filter theory, there are many good sources of that to be found on the internet. I just want to give you a small hint, an LC filter is not simply a second order 12 dB/oct filter, like you said in above post. It has a series resonance frequency, which put's the Q of the resonant circuit in focus, when you optimize the filter charactistics. As a filter designer you try to make the Q work for you, to maintain as flat and as extended response as possible, taking the load and other Q relevant components into consideration.
 
Oh, come on!
You can set the characteristic impedance to 8ohms (lets assume for now speaker is resistive up to hundreds of kHz) and Q of exactly 1. This is made by ~6uH and 100nF. You can have a RC snubber, but no but no second order filter will keep -3dB flatness up to 200kHz while attenuating the 500kHz carrier from 120Vpp to 4Vpp. We both know that.

Going back to bacics, any hi-fi amplifier (tube, solid state, class A, AB, D, X, W, ...) is demanded to have some damping factor to the load. A 10uH series inductor has ~1.25 ohms@20kHz, which means damping factor of around 6 at 8ohms and around 3 at 4 ohms, far too low. Post-filter feedback amplifiers can do much better here.

So, no need to go to OPA627 miracle-opamp and 200kHz bandwidth while the lacking is evident and well in audio range.
 
Dark Fenriz: I think we have arrived at a classic hifi dilemma here. If 0.0001% THD and a damping factor of 1000 achieved with use of large amounts of 4 point feedback was all that were required for a good audio experience, i could name 2-3 large hifi brands that were dominating in the 1980's that would now be the only surviving brands on the market.

As it is these brands are today all but extinct, while other manufacturers with more refined design parameters are doing much better.

THD matters, but it must not be achieved through the heavy use of feedback, as any electronics designer knows that sending a signal through an unlinear element, and then using feedback to linearize it, will turn the circuit into a RF mixer, creating harmonics. So it might improve the datasheet but the sound will remain distorted. We improve THD by removing dead time in the output stage, right down to 1-2 nS.

Damping factor matters too, but i think most people know that it only matters in the lower frequencies, where you want to have grip on a bass membrane. Many tweeters have resistors in series with them anyway. So that eliminates the use for high damping factor in high frequencies.
I learned very early in my hifi design life, that the only damping factor that will be audible is 'true' damping factor. In Class A/B terms this means if you have a high 'natural' output impedance, and try to improve it with feedback, the expected control of the sound will not emerge. If you remove the feedback, and replace with 'true' low impedance, i.e. placing a resistor on the 3 rd or 4th darligton set back from the output stage, then you get a good controlling effect on the sound. Even without feedback, I have sold 1000's of pure non feedback amplifiers, and the customers were very happy, many of them still are. IMO many of the paradigms from analog amplifiers can be transferred to Class D. But hey if you want to black-box the amplifier, and claim if it measures better, then it sounds better .. well ... maybe buy a nice Technics amplifier from 1985 :D

Did you ever listen to the same amplifier with pre- and post feedback to compare the sound qualities?
 
ncdxfreqpre.png


ncdxfreqpost.png


As you see the NCDX only gets to 200 kHz in the Post mode and in 8 Ohms.

So i guess i owe Darkfenriz a beer .. sorry :drink:
 
Properly impemented, the NCD is really fantastic. I have looked into the old NCD quite some time ago, and posted some impressions. However, after these years, I have discovered that if I had paid more attention to the input impedance characteristics which I have found to be cable dependent and cable length dependent, the listening impressions would have been greatly enhanced.
I had been wondering why those LDR based volume control enhances the performance to much since I had listened to one connected to the NCD. Through recent investigation going from speaker to amp in an active speaker project, I can almost be sure that the LDR based volume controls provide better impedance matching. I am also quite confident that the reason why different OP amps sound different while opinions vary, are also because of the input impedance characteristics and how it is interfaced with the other circuits.
In the last impressions posted, I was using a Hafler XL-280 amplifier that has some circuit at the input that assists in impedance problems. The NCD did not seem to have this, but this is good because the user can choose how to handle the input impedance issue based on his own cable.
Since I have tuned cables at hand, and also have some better speakers than the one originally used, I plan to go back and verify this pretty soon.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.